was noted by Gruber (1998), who estimated the
uncertainties of his calculated inventories at 20%
for exactly this reason. For example, Gruber’s
(1998) inventory of Excess CO 2 between 20 and
30°S is about 37 molC m
92 . A 2000-m thick layer
with Excess CO 2 content of 4 mol kg
91 would
contribute an additional 8 molC m
92
, or 21%,
to this inventory. This difference lies at the edge
of Gruber’s (1998) suggested uncertainty level,
thereby supporting his recommendation that ‘more
refined techniques are necessary in order to obtain
improved estimates of this error’.
This difference between the two techniques may
be related to the fact that CFC-11 is undetectable
throughout most of this depth range at these latitudes (e.g. Wallace et al., 1994; Roether and
Putzka, 1996). This may have led Gruber (1998) to
assume that this depth range was also uncontaminated with Excess CO 2 . On the other hand, both
Wallace et al. (1994) and Roether and Putzka
(1996) measured detectable levels of anthropogenic
CCl 4 within this depth range over wide regions
where CFC-11 was completely undetectable.
Anthropogenic CCl 4 has been present in the ocean
since the 1920s, at which time surface seawater
already contained about 14 mol kg
91 of Excess
CO 2 . Excess CO 2 , having been introduced since
about 1750, will have penetrated even further into
the deep ocean than CCl 4 and considerably further
than the CFCs (introduced in 1950s) and tritium
(introduced in 1960s). In the approach used by
Holfort et al. (1998), these deep CCl 4 data were
used to estimate air–sea disequilibria for deep
waters (Fig. 6.3.7) and this implied non-zero
Excess CO 2 levels between 2000 and 4000 m. The
detection of CCl 4 throughout most of the low-tomid-latitude South Atlantic (including the North
Atlantic Deep Water and parts of the eastern basin)
6.3 Storage and Transport of Excess CO 2 in the Oceans
507
Wallace
detectable in the environment around 1950 whereas CCl 4 started increasing significantly around 1910–20. The time
histories are shown up until 1992.
4 Surface water pCFC-11 (and pCCl 4 ) were assumed to have remained close to equilibrium with the atmosphere,
whereas surface water pCO 2 was expected to have deviated, proportionally, much further from equilibrium (due to
the longer air–sea equilibration time for CO 2 ).Where CFC-11 was detectable, an effective equilibrium value of the
preformed pCO 2 for the year of water mass formation (t), pCO 2
eq,t
, was estimated from the water sample’s
pCFC-11 and the reconstructed temporal equilibrium relationship between the gases (panel c). Where CFC-11
was undetectable but CCl 4 was detectable, the pCCl 4 was used for this purpose. A typical profile of pCO 2
eq,t is
presented (panel d).
5 This approach does not deal explicitly with the effect of subsurface mixing of waters with different ages. It also does
not account for non-conservative behaviour of the preformed pCO 2 (pCO 2
0
) during mixing.With respect to the
former issue, the equilibrium relationship between pCFC-11 and pCO 2
eq,t has been quasilinear over much of the
time-history of the tracer input (panel c). Hence for subsurface, along-isopycnal mixing between waters with ages
corresponding to this linear portion of the relationship, a reasonable estimate of the ‘effective’ value pCO 2
eq,t
should still result. With respect to the latter issue, a refinement of this approach based on the use of preformed
TCO 2 concentrations and linearization of the equilibrium constants linking pCO 2 and TCO 2 (cf. Gruber et al., 1996)
could be developed and implemented.
6 The difference at each depth between the measurement-derived pCO 2
0 and the tracer-derived estimate of pCO 2
eq,t
(panel d) is analogous to the air–sea disequilibrium of Gruber et al. (1996; see also equation (6.3.2)). In our South
Atlantic sections this difference was frequently positive near the surface due to seasonal effects discussed in the
text. At depth, the inferred air–sea disequilibrium reached about 45 atm, indicative of CO 2 undersaturation in
newly formed deep waters. These values are comparable to Gruber’s (1998) estimates (cf. his Fig. 7). In practice, for
all depths where either CFC-11 or CCl 4 was detectable, we expressed the air–sea disequilibria as a ratio:
pCO 2
0 /pCO 2
eq,t
. Values for this ratio based on our South Atlantic data varied from close to 1 just below the surface
layer to a minimum of about 0.85 (denoting undersaturation) at depth. Where all transient tracers including CCl 4
were undetectable, this ratio was assumed to remain constant at its lowest value.
7 These air–sea pCO 2 disequilibrium ratios were applied to the preindustrial equilibrium pCO 2 level of 280 atm in
order to calculate the preindustrial, preformed pCO 2 as a function of depth: pCO 2
0,t:preind
. (Here we assumed that
the air–sea disequilibrium has remained constant throughout the industrial era.) These estimates of pCO 2
0,t:preind
were used together with the potential temperature, salinity and the preformed alkalinity to calculate profiles of the
preindustrial preformed TCO 2 or TCO 2
0,t:preind (panel e).
8 Finally, the differences between the observation-based TCO 2
0 values (panels a and e) and the estimated
preindustrial values (TCO 2
0,t:preind
) was assumed equal to the Excess TCO 2 concentration (see also Fig. 6.3.6).
uncertainties of his calculated inventories at 20%
for exactly this reason. For example, Gruber’s
(1998) inventory of Excess CO 2 between 20 and
30°S is about 37 molC m
92 . A 2000-m thick layer
with Excess CO 2 content of 4 mol kg
91 would
contribute an additional 8 molC m
92
, or 21%,
to this inventory. This difference lies at the edge
of Gruber’s (1998) suggested uncertainty level,
thereby supporting his recommendation that ‘more
refined techniques are necessary in order to obtain
improved estimates of this error’.
This difference between the two techniques may
be related to the fact that CFC-11 is undetectable
throughout most of this depth range at these latitudes (e.g. Wallace et al., 1994; Roether and
Putzka, 1996). This may have led Gruber (1998) to
assume that this depth range was also uncontaminated with Excess CO 2 . On the other hand, both
Wallace et al. (1994) and Roether and Putzka
(1996) measured detectable levels of anthropogenic
CCl 4 within this depth range over wide regions
where CFC-11 was completely undetectable.
Anthropogenic CCl 4 has been present in the ocean
since the 1920s, at which time surface seawater
already contained about 14 mol kg
91 of Excess
CO 2 . Excess CO 2 , having been introduced since
about 1750, will have penetrated even further into
the deep ocean than CCl 4 and considerably further
than the CFCs (introduced in 1950s) and tritium
(introduced in 1960s). In the approach used by
Holfort et al. (1998), these deep CCl 4 data were
used to estimate air–sea disequilibria for deep
waters (Fig. 6.3.7) and this implied non-zero
Excess CO 2 levels between 2000 and 4000 m. The
detection of CCl 4 throughout most of the low-tomid-latitude South Atlantic (including the North
Atlantic Deep Water and parts of the eastern basin)
6.3 Storage and Transport of Excess CO 2 in the Oceans
507
Wallace
detectable in the environment around 1950 whereas CCl 4 started increasing significantly around 1910–20. The time
histories are shown up until 1992.
4 Surface water pCFC-11 (and pCCl 4 ) were assumed to have remained close to equilibrium with the atmosphere,
whereas surface water pCO 2 was expected to have deviated, proportionally, much further from equilibrium (due to
the longer air–sea equilibration time for CO 2 ).Where CFC-11 was detectable, an effective equilibrium value of the
preformed pCO 2 for the year of water mass formation (t), pCO 2
eq,t
, was estimated from the water sample’s
pCFC-11 and the reconstructed temporal equilibrium relationship between the gases (panel c). Where CFC-11
was undetectable but CCl 4 was detectable, the pCCl 4 was used for this purpose. A typical profile of pCO 2
eq,t is
presented (panel d).
5 This approach does not deal explicitly with the effect of subsurface mixing of waters with different ages. It also does
not account for non-conservative behaviour of the preformed pCO 2 (pCO 2
0
) during mixing.With respect to the
former issue, the equilibrium relationship between pCFC-11 and pCO 2
eq,t has been quasilinear over much of the
time-history of the tracer input (panel c). Hence for subsurface, along-isopycnal mixing between waters with ages
corresponding to this linear portion of the relationship, a reasonable estimate of the ‘effective’ value pCO 2
eq,t
should still result. With respect to the latter issue, a refinement of this approach based on the use of preformed
TCO 2 concentrations and linearization of the equilibrium constants linking pCO 2 and TCO 2 (cf. Gruber et al., 1996)
could be developed and implemented.
6 The difference at each depth between the measurement-derived pCO 2
0 and the tracer-derived estimate of pCO 2
eq,t
(panel d) is analogous to the air–sea disequilibrium of Gruber et al. (1996; see also equation (6.3.2)). In our South
Atlantic sections this difference was frequently positive near the surface due to seasonal effects discussed in the
text. At depth, the inferred air–sea disequilibrium reached about 45 atm, indicative of CO 2 undersaturation in
newly formed deep waters. These values are comparable to Gruber’s (1998) estimates (cf. his Fig. 7). In practice, for
all depths where either CFC-11 or CCl 4 was detectable, we expressed the air–sea disequilibria as a ratio:
pCO 2
0 /pCO 2
eq,t
. Values for this ratio based on our South Atlantic data varied from close to 1 just below the surface
layer to a minimum of about 0.85 (denoting undersaturation) at depth. Where all transient tracers including CCl 4
were undetectable, this ratio was assumed to remain constant at its lowest value.
7 These air–sea pCO 2 disequilibrium ratios were applied to the preindustrial equilibrium pCO 2 level of 280 atm in
order to calculate the preindustrial, preformed pCO 2 as a function of depth: pCO 2
0,t:preind
. (Here we assumed that
the air–sea disequilibrium has remained constant throughout the industrial era.) These estimates of pCO 2
0,t:preind
were used together with the potential temperature, salinity and the preformed alkalinity to calculate profiles of the
preindustrial preformed TCO 2 or TCO 2
0,t:preind (panel e).
8 Finally, the differences between the observation-based TCO 2
0 values (panels a and e) and the estimated
preindustrial values (TCO 2
0,t:preind
) was assumed equal to the Excess TCO 2 concentration (see also Fig. 6.3.6).